In coding of sound signals such as speech and music, a method of performing the coding by using linear prediction coefficients obtained by performing linear prediction analysis on an input sound signal is widely used.
In order to make it possible to obtain, on the part of a decoding device, the information on the linear prediction coefficients used in coding processing by decoding, a coding device codes the linear prediction coefficients and sends a code corresponding to the linear prediction coefficients to the decoding device. In Non-patent Literature 1, a coding device converts linear prediction coefficients into a sequence of LSP (Line Spectrum Pair) parameters which are parameters in a frequency domain and equivalent to the linear prediction coefficients and sends an LSP code obtained by coding the sequence of LSP parameters to a decoding device.
The outline of existing sound signal coding device 60 and decoding device 70 which are provided with linear prediction coefficient coding device and decoding device, respectively, will be described.
<Existing Coding Device 60>
The configuration of the existing coding device 60 is depicted in FIG. 1.
The coding device 60 includes a linear prediction analysis unit 61, an LSP calculation unit 62, an LSP coding unit 63, a coefficient conversion unit 64, a linear prediction analysis filter unit 65, and a residual coding unit 66. Of these units, the LSP coding unit 63 that receives LSP parameters, codes the LSP parameters, and outputs an LSP code is a linear prediction coefficient coding device.
To the coding device 60, frame-by-frame, which is a predetermined time segment, input sound signals are consecutively input, and the following processing is performed on a frame-by-frame basis. Hereinafter, specific processing of each unit will be described on the assumption that an input sound signal which is being currently processed is an fth frame. An input sound signal of an fth frame is referred to as Xf.
<Linear Prediction Analysis Unit 61>
The linear prediction analysis unit 61 receives an input sound signal Xf, obtains linear prediction coefficients af[1], af[2], . . . , af[p] (p is a prediction order) by performing linear prediction analysis on the input sound signal Xf, and outputs the linear prediction coefficients af[1], af[2], af[p]. Here, af[i] represents an ith-order linear prediction coefficient that is obtained by performing linear prediction analysis on the input sound signal Xf of the fth frame.
<LSP Calculation Unit 62>
The LSP calculation unit 62 receives the linear prediction coefficients af[1], af[2], af[p], obtains LSP (Line Spectrum Pairs) parameters θf[1], θf[2], . . . , θf[p] from the linear prediction coefficients af[1], af[2], . . . , af[p], and outputs the LSP parameters θf[1], θf[2], . . . , θf[p]. Here, θf[i] is an ith-order LSP parameter corresponding to the input sound signal Xf of the fth frame.
<LSP Coding Unit 63>
The LSP coding unit 63 receives the LSP parameters θf[1], θf[2], . . . , θf[p], codes the LSP parameters θf[1], θf[2], . . . , θf[p], obtains an LSP code CLf and quantization LSP parameters {circumflex over ( )}θf[1], {circumflex over ( )}θf[2], . . . , {circumflex over ( )}θf[p] corresponding to the LSP code, and outputs the LSP code CLf and the quantization LSP parameters {circumflex over ( )}θf[1], {circumflex over ( )}θf[2], . . . , {circumflex over ( )}θf[p]. Incidentally, the quantization LSP parameters are what are obtained by quantizing the LSP parameters. In Non-patent Literature 1, coding is performed by a method by which a weighted differential vector of the LSP parameters θf[1], θf[2], . . . , θf[p] based on a past frame is obtained, the weighted differential vector is divided into two subvectors: one on a low-order side and the other on a high-order side, and coding is performed such that each subvector becomes the sum of subvectors from two codebooks; however, there are various existing technologies as the coding method. Therefore, in coding of the LSP parameters, various well-known coding methods are sometimes adopted, such as the method described in Non-patent Literature 1, a method of performing vector quantization in multiple stages, a method of performing scalar quantization, and a method obtained by combining these methods.
<Coefficient Conversion Unit 64>
The coefficient conversion unit 64 receives the quantization LSP parameters {circumflex over ( )}θf[1], {circumflex over ( )}θf[2], . . . , {circumflex over ( )}θf[p], obtains linear prediction coefficients from the quantization LSP parameters {circumflex over ( )}θf[1], {circumflex over ( )}θf[2], . . . , {circumflex over ( )}θf[p], and outputs the linear prediction coefficients. Incidentally, since the output linear prediction coefficients correspond to quantized LSP parameters, the output linear prediction coefficients are referred to as quantization linear prediction coefficients. Here, the quantization linear prediction coefficients are assumed to be {circumflex over ( )}af[1], {circumflex over ( )}af[2], . . . , {circumflex over ( )}af[p].
<Linear Prediction Analysis Filter Unit 65>
The linear prediction analysis filter unit 65 receives the input sound signal Xf and the quantization linear prediction coefficients {circumflex over ( )}af[1], {circumflex over ( )}af[2], . . . , {circumflex over ( )}af[p], obtains a linear prediction residual signal which is a linear prediction residue by the quantization linear prediction coefficients {circumflex over ( )}af[1], {circumflex over ( )}af[2], {circumflex over ( )}af[p] of the input sound signal Xf, and outputs the linear prediction residual signal.
<Residual Coding Unit 66>
The residual coding unit 66 receives the linear prediction residual signal, obtains a residual code CRf by coding the linear prediction residual signal, and outputs the residual code CRf.
<Existing Decoding Device 70>
The configuration of the existing decoding device 70 is depicted in FIG. 2. To the decoding device 70, frame-by-frame LSP codes CLf and residual codes CRf are input, and the decoding device 70 obtains a decoded sound signal {circumflex over ( )}Xf by performing decoding processing on a frame-by-frame basis.
The decoding device 70 includes a residual decoding unit 71, an LSP decoding unit 72, a coefficient conversion unit 73, and a linear prediction synthesis filter unit 74. Of these units, the LSP decoding unit 72 that receives an LSP code, decodes the LSP code, obtains decoded LSP parameters, and outputs the decoded LSP parameters is a linear prediction coefficient decoding device.
Hereinafter, specific processing of each unit will be described on the assumption that an LSP code and a residual code on which decoding processing is being currently performed are an LSP code CLf and a residual code CRf, respectively, corresponding to an fth frame.
<Residual Decoding Unit 71>
The residual decoding unit 71 receives the residual code CRf, obtains a decoded linear prediction residual signal by decoding the residual code CRf, and outputs the decoded linear prediction residual signal.
<LSP Decoding Unit 72>
The LSP decoding unit 72 receives the LSP code CLf, obtains decoded LSP parameters {circumflex over ( )}θf[1], {circumflex over ( )}θf[2], . . . , {circumflex over ( )}θf[p] by decoding the LSP code CLf, and outputs the decoded LSP parameters {circumflex over ( )}θf[1], {circumflex over ( )}θf[2], . . . , {circumflex over ( )}θf[p]. If the LSP code CLf output from the coding device 60 is input to the decoding device 70 without error, the decoded LSP parameters obtained in the LSP decoding unit 72 are the same as the quantization LSP parameters obtained in the LSP coding unit 63 of the coding device 60.
<Coefficient Conversion Unit 73>
The coefficient conversion unit 73 receives the decoded LSP parameters {circumflex over ( )}θf[2], {circumflex over ( )}θf[p], converts the decoded LSP parameters {circumflex over ( )}θf[1], {circumflex over ( )}θf[2], . . . , {circumflex over ( )}θf[p] into linear prediction coefficients, and outputs the linear prediction coefficients. Since the output linear prediction coefficients correspond to LSP parameters obtained by decoding, the output linear prediction coefficients are referred to as decoded linear prediction coefficients and represented as {circumflex over ( )}af[1], {circumflex over ( )}af[2], . . . , {circumflex over ( )}af[p].
<Linear Prediction Synthesis Filter Unit 74>
The linear prediction synthesis filter unit 74 receives the decoded linear prediction coefficients {circumflex over ( )}af[1], {circumflex over ( )}af[2], . . . , {circumflex over ( )}af[p] and the decoded linear prediction residual signal, generates a decoded sound signal Xf by performing linear prediction synthesis on the decoded linear prediction residual signal by using the decoded linear prediction coefficients {circumflex over ( )}af[1], {circumflex over ( )}ar[2], . . . , {circumflex over ( )}ar[p], and outputs the decoded sound signal {circumflex over ( )}Xf.